Study Efficacy and the Region of Proximal Learning Framework

Nate Kornell , Janet Metcalfe

Author Affiliations

Kornell, N., & Metcalfe, J. (2006). Study efficacy and the region of proximal learning framework. Journal of Experimental Psychology: Learning, Memory, and Cognition, 32(3), 609-622. doi:10.1037/0278-7393.32.3.609

Abstract

One of the most important reasons to investigate human metacognition is its role in directing how people study. However, limited evidence exists that metacognitively guided study benefits learning. Three experiments are presented that provide evidence for this link. In Experiment 1, participants’ learning was enhanced when they were allowed to control what they studied. Experiments 2a–d replicated this finding and showed contributions of self-regulated study to learning. Experiments 3a and 3b showed that, when forced to choose among items they did not know, participants chose the easiest items and benefited from doing so, providing evidence for the link between metacognitive monitoring/control and learning, and supporting the region of proximal learning model of study-time allocation.

KEYWORDS:

metacognition , study-time allocation , region of proximal learning , education , memory

One of the best reasons to study metacognition is because it has the potential to play a large part in guiding how people study and, as a result, in how effectively they learn (e.g., Benjamin, Bjork, & Schwartz, 1998; Dunlosky & Hertzog, 1997; Metcalfe, 2002; Metcalfe & Kornell, 2003, 2005; T. O. Nelson & Narens, 1990, 1994). People use memory monitoring, especially judgments of learning (JOLs), to decide which items to study and how long to spend on them (e.g., Mazzoni, Cornoldi, & Marchitelli, 1990; Metcalfe, 2002; T. O. Nelson, Dunlosky, Graf, & Narens, 1994; T. O. Nelson & Leonesio, 1988; T. O. Nelson & Narens, 1990; Son & Metcalfe, 2000). The central question addressed here is, does such metacognitively guided study lead to effective learning? The answer can be arrived at by asking more basic questions. What should people choose to study? What do people choose to study? And, are they the same? The answers to these questions have implications for pedagogy, how learning works, and the debate over the accuracy (or lack thereof) of metacognitive monitoring and control as well as theoretical implications for current models of study-time allocation.

Until recently, the dominant model of study-time allocation has been discrepancy reduction (DR) (e.g., Dunlosky & Hertzog, 1998). According to this model, a person chooses to study the items that are farthest from being learned, that is, the most difficult items for that person (e.g., Dunlosky & Hertzog, 1998; Dunlosky & Thiede, 1998; T. O. Nelson & Narens, 1990; Thiede, Anderson, & Therriault, 2003; Thiede & Dunlosky, 1999; note that Dunlosky & Thiede, 2004, previously proponents of this position, have changed their views.) DR also posits that directing study to the most difficult items is the most effective study strategy, and a negative correlation between JOLs and study-time allocation, which this model predicts, has been taken to indicate an effective study-time allocation strategy. Focusing study where one is least competent seems intuitive (see Woodworth, 1921), and many studies have shown that people preferentially study the most difficult items (for a review, see Son & Metcalfe, 2000). These results provide support for DR, with one caveat: The experiments were all limited to situations in which it made sense for participants to try to master all of the materials because the time allowed for study was unlimited, so studying one item for longer did not mean another would be given short shrift, and because the items were all within a limited range of difficulty. It stands to reason that, to achieve mastery, the most difficult items must be studied most.

But total mastery is not always possible. In real life, time constraints often exist that prevent the learner from spending enough time on difficult items to master them. Under time pressure or when the performance goals are lower than complete mastery (see Dunlosky & Thiede, 2004), it may be suboptimal to emphasize items that are too difficult to learn. Indeed, T. O. Nelson and Leonesio (1988) have demonstrated that devoting excessive time to the most difficult items may not help learning, an effect that they dubbed the “labor-in-vain” effect. DR breaks down when people are given limited time or an easily attainable performance goal, when choosing one item means sacrificing another, and when people are faced with selecting only some of the items they do not know (Metcalfe & Kornell, 2005). When the only factor under consideration is choice of which items to study, as is the case in all of the experiments in the present article, then even proponents of DR (Dunlosky & Thiede, 1998, 2004), as well as those who argue that it is fatally flawed (e.g., Metcalfe & Kornell, 2005), have proposed alternative models.

According to an alternative, the region of proximal learning (RPL) framework (Metcalfe, 2002; Metcalfe & Kornell, 2003, 2005), people focus on studying the easiest items they do not know, prioritizing easier unknown items over the most difficult, especially when all of the unknown items cannot be studied. Prioritizing moderately difficult items is the most effective strategy according to RPL. When it is possible to try to master the entire set of to-be-learned items, RPL and DR make similar predictions because learning the difficult items will require the most study time. Indeed, DR reduces to a specific case of RPL. The difference occurs when people perceive that they have little or no chance of learning the most difficult items, given the constraints of the situation. Under such conditions, they will stop spending time on the most difficult items, instead turning to easier items that they believe they can learn.

In addition to predicting what people should and do study, RPL also identifies the processes by which study decisions are made. According to RPL, study-time allocation has two separable components: choice of items to study and perseverance on items once they are selected (Metcalfe & Kornell, 2005). Study choices are made by selecting the easiest items one does not yet know and, among them, prioritizing those that are closest to being learned. Once a choice has been made, persistence depends on the perceived rate of information uptake, a measure that Metcalfe and Kornell called a judgment of the rate of learning (jROL). People are posited to persist until they perceive that they have ceased learning.

The evidence suggests that, when people choose what to study, they eliminate what they already know (e.g., Cull & Zechmeister, 1994; Masur, McIntyre, & Flavell, 1973; Son, 2004). But what happens next, when people chose among items they do not know (and when DR and RPL make opposite predictions)? Because almost all study-time allocation studies show that people select difficult items, one might expect people to select the most difficult of the unknown items, contradicting RPL. However, the data show no such contradiction. Metcalfe and Kornell (2005) reviewed the extant literature on study choice and concluded that “among the choice studies we have been able to find, even those showing very large negative gamma correlations, none have demonstrated that people preferentially chose the least learned items among those that were not already learned” (p. 470). Instead, the negative correlations appear to be due to the simple fact that, during the first phase of choice, people do not choose to study what they already know.

Since Son and Metcalfe’s (2000) literature review, which showed what was at the time a pervasive negative correlation between study-time allocation and JOLs, new evidence has come to light supporting RPL. Son and Metcalfe’s own data showed that, when total time studying was limited and thus more time studying one item meant less was available for others, people focused on easy items, supporting RPL and contradicting DR. Thiede and Dunlosky (1999) found a similar effect. Metcalfe (2002) and Metcalfe and Kornell (2003) replicated this finding using a different paradigm and showed that people focused on moderately difficult items. These latter studies also showed that, when more time was allowed for study, people increasingly focused on difficult items, which RPL predicts for two reasons. First, as time passes and learning progresses, the set of almost-but-not-quite-learned items (i.e., items in their region of proximal learning) shifts to more difficult items. Second, as study time increases, the possibility of mastery increases, and people become more inclined to try to learn the most difficult items. Similarly, Metcalfe also found, using Spanish vocabulary pairs, that Spanish speakers studied more difficult materials than did non-Spanish speakers, although not the most difficult, again zeroing in on their own region of proximal learning. Finally, people shift to studying easier items when given goals that do not require learning all items (Dunlosky & Thiede, 2004; Thiede & Dunlosky, 1999).

These studies all suggest that assessing the goodness of metacognitive control in terms of the strength of the negative correlation between metacognition and study choice may be inappropriate. Instead, people seem to choose to study items that are in their region of proximal learning. However, not one of these studies has demonstrated that this strategy is effective. Indeed, despite its importance, the evidence for an effective monitoring–study–learning link is tenuous (T. O. Nelson et al., 1994; Thiede, 1999; Thiede et al., 2003).

Previous Findings on the Effectiveness of Metacognitively Guided Study

Making a link between metacognition and learning requires two skills on the part of the learner: monitoring learning and memory, and controlling study effectively based on that monitoring (T. O. Nelson & Narens, 1990). Thus, effective study involves linking…